CD spectroscopy reveals structural changes

Circular dichroism is a powerful tool to monitor structural changes in proteins as well as in nucleic acids. Regular A-form RNA has a characteristic CD spectrum showing a large positive elipticity at 260 nm and a negative one at 210 nm.(388) Changes in the characteristic band at 260 nm correspond to hyperchromism of nucleobases as a result of helical stacking(389) and the band at 210 nm is associated with changes in loop structures.(296) Thus, changes in the shape of the CD spectrum can be directly related to structural changes in the RNA.

A comparison of the CD spectra of the wildtype d3'-EBS1 and d3'-EBS1·IBS1 (Figure 108A) show solely a general increase in intensity, i.e. the shape of the CD curve does not change, indicating that no structural change occurs. This is explained by the additional presence of nucleobases from IBS1, i.e. an increase of hyperchromism is seen. This finding indicates that IBS1 does not bind to EBS1 under these conditions being in agreement with UV melting studies described in Section 2.2.3.

d3'-EBS1* and d3'-EBS1*·IBS1* behave differently: Addition of the intronic sequence to d3'-EBS1* leads to a slight change in elipticity as can be seen best by the decreased intensity of the shoulder at 280 nm (compare solid lines in Figure 108B and C). This decrease is well in line with an expected different backbone structure in d3'-EBS1* compared to d3'-EBS1*·IBS1*, similar to previous observations with other RNAs.(296,389) Melting of the two RNAs has a twofold effect on both CD spectra: At room temperature, the CD signal of both d3'-EBS1* and d3'-EBS1*·IBS1* displays the characteristic shape of an A-form RNA together with a more distinct additional shoulder at

Figure 108 (A) Change in the CD signal of d3'-

EBS1 (solid line) upon addition of 1 eq IBS1 (broken line) in the presence of 100 mM KCl. (B) Melting of d3'-EBS1* as observed with CD spectra in the presence of 10 mM KCl. (C) Change in CD spectra of d3'-EBS1*·IBS1* with increasing temperature in the presence of 100 mM KCl. The concentrations of the RNA are 10 µM (A) and 16 µM (B, C)

280 nm in the case of d3'-EBS1*·IBS1*. Upon heating to 60 °C, the elipticity at 210 nm and 260 nm decreases in both cases. However, whereas the CD signal at 280 nm of d3'-EBS1* looses intensity, in the case of the complex d3'-EBS1*-IBS1*, the opposite trend is observed giving an isosbestic point at 270 nm. A further increase in temperature to 80 °C then leads to a drastic change of the CD spectra as exhibited by a red-shift of the largest positive elipticity, giving rise to final spectra as expected for unstructured RNAs.

The above described observations are well in line with the formation of the d3'- EBS1*·IBS1* complex: The first melting transition (see also Table 4) corresponds to the dissociation of IBS1* yielding a spectral trace dominated by the still partly structured d3'- EBS1*. The hairpin stem of d3'-EBS1* then melts at higher temperature giving rise to the observed red shift and the new maximum elipticity at 280 nm.

In order to investigate the influence of Mg2+ on the formation of the recognition site complex, binding of IBS1* to d3'-EBS1* was monitored by CD spectroscopy under different conditions. The following three modes of actions can be envisaged: (i) Mg2+ is needed to pre- structure the EBS1* loop such that IBS1* can bind efficiently, (ii) Mg2+ is not needed for

Figure 109 Changes in CD spectra of d3'-EBS1* (black solid line) upon addition of 1 eq IBS1* (dark blue

dashed), 20 eq Mg2+ or Cd2+, respectively (cyan dot-dashed) and 1 eq IBS1*/20 eq Mg2+ or Cd2+, respectively

(red dotted) in the presence of 10 mM KCl (A and C) and 100 mM KCl (B and D). Three isosbestic points at 205

nm, 245 nm and 270 nm (circles) are observed upon addition of IBS1* and IBS1*/20eq Mg2+ or Cd2+,

Results and Discussion 153 binding of IBS1* to EBS1*, but for the final structure of the 5'-splice site, or (iii) Mg2+ is not needed at all. Hence, we recorded CD spectra between 200 nm and 320 nm of d3'-EBS1* alone, after the addition of one equivalent of IBS1*, or 20 equivalents of Mg2+, respectively, and in the presence of both, i.e. one equivalent IBS1* and 20 equivalents Mg2+ (Figure 109A and B). Spectra were recorded at a background concentration of either 10 mM or 100 mM KCl.

In the presence of 10 mM KCl the addition of IBS1* to d3'-EBS1* leads to an increase in elipticity at 210 nm and 260 nm (Figure 109A). This increase is a consequence of binding of IBS1* to its cognate partner being consistent with the increasing number of stacked nucleobases upon addition of IBS1*. Subsequent addition of micromolar amounts of Mg2+ further increases significantly both bands suggesting that Mg2+ is needed to form the fully stable d3'-EBS1*·IBS1* complex (Figure 109A). The presence of higher concentrations of Mg2+ does not lead to any further changes in the CD spectrum. To investigate if the order of addition of Mg2+ or IBS1* to d3'-EBS1* has any influence on the final structure, a second experiment was performed, in which Mg2+ was added prior to IBS1*. These experiments show that the order of addition has no influence on the final structure. In the absence of IBS1*, the addition of Mg2+ leads to an intensification of the shoulder at around 280 nm and a slight shift to higher wavelength. Obviously Mg2+ binds to d3'-EBS1* inducing a structural change. The increasing elipticity at 280 nm suggests a longer single stranded, i.e. more unstructured region, similar to the melting of a double helix (see also Figure 108B and C). Indeed, the loop including EBS1* can form further basepairs, i.e. one Watson-Crick and two wobble base pairs, extending the duplex region and capped by a tetraloop (Figure 39B). Estimates of the folding energies of the open loop and the extended duplex region, respectively,(26,310) show that both forms have almost identical stabilities. The coordination of Mg2+ to d3'-EBS1* could thus lead to the melting of the EBS1* region thereby facilitating subsequent binding of IBS1*. Mg2+ binding to this region and EBS1* to be single stranded in the presence of Mg2+ is well in line with previously described Tb3+ cleavage experiments, where it has been shown that not only a metal ion binding site is located just 5' of EBS1 (also in the absence of IBS1), but in addition that EBS1 is not part of a duplex in the absence of 5'- exon.(91,97)

To investigate if Mg2+ is required to sculpt the 5'-splice site or if also monovalent ions can stabilize such a structure, the corresponding experiments were performed in the presence of 100 mM KCl (Figure 109B). As is the case in the presence of 10 mM KCl, also in 100 mM, the addition of IBS1* to d3'-EBS1* leads to an absolute increase in elipticity at 210 nm and

260 nm. However, at high K+ concentration, the subsequent addition of Mg2+ does not result in a further increase in intensity. This suggests that at high concentration of K+, binding of IBS1* is already fully achieved. Nevertheless, the small change in intensity at 280 nm indicates that Mg2+ still binds and has some influence on local structure, most probably in the unpaired region 5' of EBS1*. It is interesting to note that the shoulders at 280 nm of the d3'- EBS1* spectra in the absence and presence of Mg2+ show a similar elipticity. The addition of IBS1* reduces the same in both instances, which is different from the situation at 10 mM KCl. Taken together, the above results show that the presence of either Mg2+ or large excess of K+ is sufficient to melt the EBS1* region prior to base pairing with IBS1* and to allow folding to the (almost) final d3'-EBS1*·IBS1* complex.

The group II intron Sc.ai5γ is known to be very sensitive to the presence of divalent metal ions other than Mg2+.(70,390) Here, we investigated how Cd2+ binds to the 5'-splice site, as this metal ion is often used in so-called thiorescue experiments,(64,95,219,243) and has the least influence on splicing.(390) The changes of the CD spectra upon addition of Cd2+ to d3'-EBS1* or d3'-EBS1*·IBS1*, respectively, are shown in Figure 109C and D. At low K+ concentration, Cd2+ again supports the formation of the d3'-EBS1*·IBS1* complex. Comparison of the overlay of CD spectra (Figure 109A and C) shows one major difference at 280 nm, i.e. Cd2+ has no influence on the intensity at this wavelength suggesting that it does bind differently to d3'-EBS1* compared to Mg2+ and does not promote melting of the EBS1* region. In the presence of 100 mM KCl the addition of Cd2+ to either d3'-EBS1* or d3'-EBS1*·IBS1* has no effect at 260 nm and 280 nm, but only at 210 nm, which is not observed in the case of Mg2+. Again this suggests that Cd2+ binds slightly different than Mg2+ inducing a slightly different structure.

To summarize, the requirement of metal ions on the formation of the 5'-splice site have been investigated. Detailed circular dichroism studies showed that metal ions are required to form the stable EBS1*·IBS1* interaction. This can be achieved by either micromolar amounts of Mg2+ or high millimolar concentrations of K+, although it seems as if Mg2+ is still needed for the final local structure. The role of these ions is most probably twofold: Firstly, Mg2+ (or 100 mM K+) leads to an increased content of single strand in d3'-EBS1* suggesting the full melting of the EBS1* region. Obviously, this facilitates the base pairing with its cognate IBS1* sequence and hence also splicing in general. Secondly, small changes in the CD spectra upon addition of Mg2+ to d3'-EBS1*·IBS1* at both 10 and 100 mM KCl suggest minor structural changes in the RNA dimer. The most likely coordination site for metal ions in the complete 5'-splice site are the unpaired nucleotides just 5' of EBS1. Indeed, Tb3+

Results and Discussion 155 cleavage experiments on the full length Sc.ai5γ are well in line with our results as they revealed EBS1 to be single stranded in the absence of IBS1 in addition to a strong metal ion binding site 5' of EBS1.(91,97)

All group II intron ribozymes investigated to date have a strict requirement for divalent metal ions for folding and catalysis.(129,203,207,243,287,391,392) In the case of Sc.ai5γ, a recent study showed that this group II intron is highly specific for Mg2+, as the partial substitution of Mg2+ with other divalent ions like Ca2+, Mn2+, or Cd2+ leads in all cases to a significant loss in activity.(390) For example, in the presence of 90 mM Mg2+ and 10 mM Cd2+, only about 75% of product was observed after 40 min compared to the reaction in 100 mM Mg2+.(390) The here observed minor differences in Mg2+ and Cd2+ binding to d3'-EBS1*·IBS1* support this high selectivity for one kind of metal ion, as every small change in local structure will have an effect on the catalytic rate. However, the fact that Mg2+ can be (mostly) replaced by high concentrations of K+ in the d3'-EBS1*·IBS1* complex is well in line with similar results from other small RNAs. For example, the hammerhead, hairpin, and VS ribozymes are still catalytically active at molar concentrations of M+ ions (in the absence of Mg2+), although at a significantly reduced rate.(74) The substitution of Mg2+ by large amounts of M+ ions in small nucleotide systems is well known: The dephosphorylation of adenosine 5'-triphosphate works best in the presence of one equivalent each of Mg2+ and Zn2+.(393) Mg2+ thereby only binds to phosphate oxygens whereas Zn2+ bridges phosphates and the purine N7 position. In the ATP system, the Mg2+ ion can be replaced by a 500-fold excess of Na+, without resulting in a loss in dephosphorylation. It thus seems only reasonable that in the here investigated d3'- EBS1*·IBS1* system a 5000-fold excess of K+ compared to Mg2+ is sufficient to induce a highly similar if not the same three dimensional local structure of the 5'-splice site.